Configure data retention for the sources of the load_streams system table ..

marcothesane

Hi All -

I'm trying to ensure that the load_streams system table has 3 days' worth of data at all times.

I just tried:

SELECT GET_DATA_COLLECTOR_TIME_POLICY(
  'v_monitor.load_streams'
, '3 days'::INTERVAL
);

Then, as this is the only 3-days' setting I ever did on this new database, I tried:

select 
  component
 ,interval_time
 , memory_buffer_size_kb
 , disk_size_kb 
from v_monitor.data_collector 
where interval_time='3 00:00:00.000000;

And got:

component               |interval_time    |memory_buffer_size_kb|disk_size_kb
Errors                  |3 00:00:00.000000|                1,000|      10,000
ExecutionEngineProfiles |3 00:00:00.000000|                2,000|      50,000
ExecutionEngineResources|3 00:00:00.000000|                2,000|      50,000
ProjectionsUsed         |3 00:00:00.000000|                2,000|     150,000
RequestsCompleted       |3 00:00:00.000000|                2,000|      50,000
RequestsIssued          |3 00:00:00.000000|                2,000|      50,000

Question now is: Do I need to modify memory_buffer_size_kb and/or disk_size_kb to make sure I get 3 days' worth of data, even if I get more data than those values state, or does the 3 day interval simply override everything?

Bryan_H

Hi Marco, according to docs, "If you specify interval, Vertica enforces the setting that is exceeded first (size or time). Before you include a time restraint, verify that the disk size capacity is sufficiently large."
See https://www.vertica.com/docs/9.2.x/HTML/Content/Authoring/SQLReferenceManual/Functions/VerticaFunctions/DataCollection/SET_DATA_COLLECTOR_POLICY.htm